Are you using mISDN 1_1_x or the git master branch ?
in mISDN_1_1 you cannot choose which of the cards is used first, in the
git master branch this is possible, by just defining the E1 card as
first device in the mISDN.conf file, the hfcmulti driver will configure
the chips in this order instead of the PCI order. Also the
SINGLE_INTERRUPT define is only available in the git master branch,
additionally the SINGLE_INTERRUPT define is not defined by default which
means you don't need to undef it at all.
The SINGLE_INTERRUPT define is also experimental, you can try to enable
it to reduce the Interrupt load of a System with a lot of HFC Chips.
I would not suggest to use the current master branch of mISDN or
mISDNuser in a production environment, instead i would suggest to use
mISDN_1_1_2 and mISDNuser_1_1_2.
These should be used with the misdn-init script which reads the
misdn-init.conf file located in /etc/misdn-init.conf.
There you should use the following settings with your setup:
card=1,0x8,pcm_slave,ignore_pcm_frameclock
card=2,0x1,
card=3,0x1,pcm_slave
This is a rather confusing configuration, but until mISDN_1_2 is
released this is the way you should go for production.
If you want to test the cool new features of the git master branch you
are indeed encouraged to do so, please let us know about your findings,
we're quite eager to solve issues, and are intersted in any opinion.
cheers,
Christian
Frank Sautter wrote:
i hope this is the right list, because i wasn't able to find a special
mISDN-dev mailing list.
we have a setup of one 8S0 and one 2E1 card which are interconnected by
a PCM cable. one E1 port is connected to the provider, the other E1 to
our old ericsson pbx, the S0 port are connected to some internal a/b
adaptors and another interconnect to our ericsson (no connection to a
telco provider on the S0 ports)
therefore our clocksource for all cards has to be the E1 coming from our
telco.
the problem is, that because the two E1 chips are behind a pci-bridge
(on the 2E1 card), they are the last devices in the pci-device-list and
the 8S0 is the first device which is recognised by the hfc_multi kernel
module, no matter in which order the cards are put into the pci slots.
but, being the first device and having no clocksource the 8S0 card lets
fail the hfc_multi kernel-module during initialisation.
the problem is, that this piece of code in hfc_multi.c overwrites any
settings of 'ignore_pcm_frameclock="yes"' in mISDN.conf without any warning.
#ifdef SINGLE_INTERRUPT
/*
* only first device gets timer interrupt, so the
* others don't have to wait for it
*/
if (HFC_idx > 0)
test_and_set_bit(HFC_CHIP_CLOCK_IGNORE, &hc->chip);
#else
if (type & 0x1000)
test_and_set_bit(HFC_CHIP_CLOCK_IGNORE, &hc->chip);
#endif
i don't have any idea, but to undefining SINGLE_INTERRUPT (with all it's
disadvantages) or to only enable the else part of this code fragment
(but this might enable more then one interrupt).
i think there should be an easy possibility in mISDN to make an incoming
E1 link to the PCM and clock master of all connected ISDN hardware.
any comments / code changes are welcome!
regards
frank sautter
_______________________________________________
Misdn-asterisk mailing list
[email protected]
http://lists.beronet.com/cgi-bin/mailman/listinfo/misdn-asterisk
_______________________________________________
Misdn-asterisk mailing list
[email protected]
http://lists.beronet.com/cgi-bin/mailman/listinfo/misdn-asterisk
